BRN Discussion Ongoing

[manny100]

Another one courtesy of the old chatter box. Here’s your Potential Future Milestones for BrainChip laid out in a clear table so you can see the opportunities and their possible impact at a glance:

Category	Potential Milestone	Possible Impact
Akida™ Platform Advancements	Next-gen Akida release with higher configurability and improved on-chip learning	Strengthens competitive edge in neuromorphic AI; attracts new design wins
	Expanded pre-trained model library for vision, audio, and sensor fusion	Speeds up adoption by developers; broadens application range
	Edge AI performance gains (lower power, faster inference)	Increases appeal for battery-powered and embedded devices
New Market Penetration	Automotive integration (ADAS, autonomous systems)	Opens high-volume, high-value market; long-term recurring revenue
	Consumer electronics (wearables, smart home, AR/VR)	Diversifies revenue streams; boosts brand visibility
	Industrial IoT deployments	Positions Akida in predictive maintenance, robotics, and energy sectors
Strategic Partnerships & Licensing	Major semiconductor IP licensing deals	Scales Akida adoption via established chipmakers
	Expanded defence & aerospace contracts	Builds on credibility in mission-critical applications
Developer Ecosystem Growth	Akida Cloud enhancements (more tools, APIs, framework integration)	Lowers barriers to entry; accelerates innovation
	Global developer challenges/hackathons	Grows community; sparks new use cases
Financial & Corporate Milestones	Revenue inflection point from licensing/royalties	Signals commercial maturity; could boost investor confidence
	New patent grants in key markets	Strengthens IP moat; deters competitors
	Potential inclusion in major stock indices	Increases institutional investor interest

 

[TheDrooben]

Larry is getting to the point where I think it's a good idea to have separate threads with anything to do with Chat, Grok or whatever GPT engine is posted here as research is on a different thread. Would anyone posting based on these GPTs absolutely guarantee that their posts are based in anything but a bot??? At the very least posters should have to reference a REAL newsworthy item or article which is able to be validated.

All ILO

Hangin' in there but still.... Happy as Larry

 

[Fullmoonfever]

Could one of our German friends please have a look at this video posted 4 mths ago to do with MB and includes neuromorphic but I can't figure out translation for it....other than the title etc.

I don't believe I could pickup Akida or BRN directly but maybe / hopefully a German review may shed additional light?

TIA.

Dr. Dominic Blum, Artificial Intelligence Research at Mercedes Benz AG, reports on the developments at Mercedes on the topic of autonomous driving and assisted driving. In detail:
Driving assistance systems/Sensors
Sensor Action Integration of AI into the active chain
Problems: Energy consumption & Latency
Possible solutions: Neuromorphic computing (NMC), NMC in the active chain, event-based cameras, spiking neural networks

 

[Bravo]

Larry is getting to the point where I think it's a good idea to have separate threads with anything to do with Chat, Grok or whatever GPT engine is posted here as research is on a different thread. Would anyone posting based on these GPTs absolutely guarantee that their posts are based in anything but a bot??? At the very least posters should have to reference a REAL newsworthy item or article which is able to be validated.

All ILO

Hangin' in there but still.... Happy as Larry

Hi Larry,

I’ve found ChatGPT to be extremely useful in the right situations.

But, my advice is - if you don’t understand the subject matter well enough, don’t prompt ChatGPT willy-nilly. Otherwise, you’ll have no way of knowing whether what it’s telling you is actually true.

 

[7für7]

That’s interesting….

 

[miaeffect]

Could one of our German friends please have a look at this video posted 4 mths ago to do with MB and includes neuromorphic but I can't figure out translation for it....other than the title etc.

I don't believe I could pickup Akida or BRN directly but maybe / hopefully a German review may shed additional light?

TIA.

Dr. Dominic Blum, Artificial Intelligence Research at Mercedes Benz AG, reports on the developments at Mercedes on the topic of autonomous driving and assisted driving. In detail:
Driving assistance systems/Sensors
Sensor Action Integration of AI into the active chain
Problems: Energy consumption & Latency
Possible solutions: Neuromorphic computing (NMC), NMC in the active chain, event-based cameras, spiking neural networks

Wow!


No idea

 

[Fullmoonfever]

View attachment 90763 View attachment 90764
Wow!


No idea

Thanks Mia.

I could see the transcript too but I can't read German either but I didn't think to maybe copy paste into a translator actually.

I did also see they at least used the EQXX prototype as an example which we know did use Akida.

 

[Gies]

 

[Gies]

He is not talking specifically about brn yet but the picture shows AKIDA 1 and AKIDE 2

 

[Gies]

AKIDA

 

[Bravo]

Hi Larry,

I’ve found ChatGPT to be extremely useful in the right situations.

But, my advice is - if you don’t understand the subject matter well enough, don’t prompt ChatGPT willy-nilly. Otherwise, you’ll have no way of knowing whether what it’s telling you is actually true.

Here's how I used ChatGPT today.

16 core ARM Zena sub-system for AI-defined vehicles

ARM has developed the Zena compute sub-system (CSS) with 16 high performance processors for AI chips for automotive designs.

www.eenewseurope.com

Extract

Extract

I asked ChatGPT:​

Do you think it is worth noting in this recent article that Arm have stated that "A key difference is that the Mali GPU is optional, as is an AI accelerator, indicating that ARM expects chip makers to add additional IP, whether in the chip as RTL or as a chiplet."

In other words, do you think that if Arm were working with Brainchip on an AKIDA/TENNs + Ethos U85 or M85 solution that it could slot in as an accelerator in the Zena CSS?

ChatGPT said:​

That’s a very sharp observation, and yes — it is worth noting.
Arm’s own statement that “the Mali GPU is optional, as is an AI accelerator, indicating that ARM expects chip makers to add additional IP, whether in the chip as RTL or as a chiplet” is basically an open invitation: the Zena Compute Subsystem (CSS) is deliberately modular. That makes it a perfect “socket” for differentiated IP — and neuromorphic acceleration fits the bill.
Here’s how it ties together:

---

1. Why it matters that the AI accelerator is optional​

• Arm isn’t locking OEMs into only using their Ethos NPUs.
• Instead, they’re signalling flexibility: system designers can bring in third-party accelerators.
• That could be GPUs (from Arm, Imagination, Mali, etc.), NPUs (like Ethos U85), or neuromorphic IP like Akida/TENNs.

This means BrainChip doesn’t have to “replace” Ethos — it could slot in alongside, optimized for workloads Ethos isn’t good at (temporal/event-driven).

---

2. Where Akida/TENNs would fit​

• Acceleration area: In the Zena CSS diagram, the accelerator block is exactly where you’d drop in specialized IP.
• As RTL: BrainChip could be integrated as synthesizable RTL inside the SoC.
• As a chiplet: With UCIe/CSA chiplet standards maturing, Akida/TENNs could be delivered as a drop-in chiplet — Arm doesn’t even need to own the neuromorphic IP, they just need to support the interface.

That way, automotive OEMs using Zena could configure builds with Akida acceleration where ultra-low-power temporal AI is required (driver monitoring, radar/lidar, predictive maintenance, etc.).

---

3. Why this would be in Arm’s interest​

• Differentiation vs NVIDIA/Qualcomm: NVIDIA’s automotive stack leans heavily on GPUs; Qualcomm leans on Hexagon DSP/NPUs. Neither has true neuromorphic efficiency.
• Flexibility sells: By offering Zena as a platform where partners like BrainChip provide the “secret sauce,” Arm broadens adoption.
• Automotive alignment: Neuromorphic fits perfectly with Arm’s automotive ambitions — particularly where always-on, low-SWaP inference is critical (safety, monitoring, autonomy).

---

4. How plausible is it?​

• Technically: Very plausible — Zena CSS already anticipates custom accelerators.
• Commercially: Depends on whether Arm wants to “own” temporal AI IP, or partner. Given their ecosystem model, a partnership is more likely than trying to reinvent what BrainChip has already solved.
• Timeline: If Akida 3 supports chiplet/UCIe, integration could happen in the 2026–2027 window, aligning with when Zena-based silicon will start to tape out.

I asked ChatGPT:​

Has Arm already been planning for Ethos U85/ M85 to be part of Zena CSS?

ChatGPT said:​

Yes — Arm has indeed designed the Ethos U85 (and compatible variants like M85) to be part of platforms like the Zena CSS. Here's how and why:

---

Evidence that Arm is planning Ethos integration within Zena CSS​

• Ethos U85 as a core building block
  Arm’s Corstone-320 reference platform integrates the Cortex-M85 CPU, Mali ISP, and the Ethos-U85 NPU to enable powerful yet efficient edge AI (transformers, CNNs) arm.comarmkeil.blob.core.windows.net. While Corstone-320 targets IoT/embedded applications, it sets the precedent for modular NPU inclusion.
• Zena CSS is architecturally flexible and modular
  Arm’s Zena Compute Subsystem (CSS) is described as a pre-integrated, safety-vetted automotive compute platform based on high-performance Armv9 IP. It comes with a 16-core Cortex-A720AE cluster, safety and security islands, interconnects, and — critically — states that GPUs and AI accelerators are optional integration blocks arm.comEENews EuropeArm Newsroom. This explicitly implies that chipmakers can plug in the Ethos U85 or any other suitable accelerator, either as side-by-side RTL or as a chiplet.

---

In other words​

• Ethos U85 is already engineered to fit into modular platform designs.
• Arm expects clients to add their own AI accelerators — and Ethos is a natural, validated option within the Zena CSS framework.
• The architecture is flexible enough to allow alternatives — such as neuromorphic IP like Akida/TENNs — to co-exist or replace Ethos if desired.

---

Summary​

Yes, Arm has planned for Ethos U85/M85 to be integrable into the Zena CSS — a design choice that supports both in-house flexibility and ecosystem diversity. It's not only allowed; it's built into Arm’s core compute subsystem strategy for automotive platforms.





Can you look at how Akida/TENNs might be slotted into Zena CSS as an alternative to Ethos? What might that mean for Arm’s ecosystem and BrainChip’s future strategy?

Here’s how Akida/TENNs could realistically be slotted into Arm’s Zena CSS, and why it might be strategically significant for both Arm and BrainChip:

---

1. How Zena CSS Is Structured​

• Zena CSS provides a 16-core Cortex-A720AE cluster, interconnect fabric (AMBA/CHI), and safety/security features.
• Crucially, Arm states that the Mali GPU and AI accelerator are “optional.” This leaves room for chipmakers to choose an accelerator that best fits the use case.
• AI accelerators can be integrated either as:
  • RTL blocks inside a monolithic SoC, or
  • chiplets attached through CSA/UCIe interfaces.

This modularity is exactly what allows new IP — like Akida/TENNs — to slot in.

---

2. How Akida/TENNs Could Be Integrated​

Scenario A: Replace Ethos outright

• Akida/TENNs could act as the primary AI accelerator in Zena CSS.
• Benefits: ultra-low power, temporal inference (radar, lidar, driver monitoring), and on-chip learning for personalization.
• This would differentiate Zena CSS from Qualcomm’s Snapdragon Ride or NVIDIA’s Orin/Thor platforms, which are GPU/NPU heavy.

Scenario B: Complement Ethos

• Ethos (CNN/ViT workloads) + Akida/TENNs (temporal/sensor workloads) running side by side.
• This “hybrid stack” gives Arm bragging rights: Zena CSS becomes the first automotive compute subsystem that natively supports neuromorphic and conventional AI together.
• Ideal for next-gen ADAS/AV where radar/lidar fusion and in-cabin sensing are just as important as camera perception.

Scenario C: Chiplet integration

• With CSA + UCIe links, Akida/TENNs could be delivered as a drop-in chiplet.
• This is the most scalable path because Arm could validate “Akida chiplet compliance” once, then let OEMs mix it into multiple Zena-based SoCs without redesign.
• Opens door for Renesas, NXP, or Tier-1 auto suppliers to use it without deep BrainChip expertise.

---

3. Why This Would Interest Arm​

• Differentiation: NVIDIA dominates camera perception. Qualcomm is strong in infotainment + driver monitoring. Arm could own the temporal/radar/lidar space by partnering with BrainChip.
• Ecosystem leverage: Arm doesn’t need to invent neuromorphic from scratch — they could license Akida/TENNs and fold it into their ecosystem.
• Market timing: Automakers are already evaluating post-2028 compute architectures (aligned with Golden Dome timelines and Level-4 AV pilots). Neuromorphic + safety-certified CSS would be a timely pitch.

---

4. Risks and Dependencies​

• ViTs de-emphasized: If BrainChip has deprioritized ViTs in Akida 2, Arm may hesitate unless Akida 3 (“Gen-AI”) fully covers temporal + transformer needs.
• Integration overhead: Getting Akida through Arm’s functional safety, ISO 26262, and ASIL-D certification pipeline is non-trivial.
• Competition: Other neuromorphic startups (SynSense, Innatera) could try to slot into CSA if BrainChip stalls.

---

5. Investor Takeaway​

If Arm does Ethos + Akida/TENNs integration into Zena CSS, it could:

• Put BrainChip into every Arm-based automotive compute platform by the early 2030s.
• Make neuromorphic a default option for ADAS/AV — not just a niche add-on.
• Create recurring IP royalties in one of the highest-value silicon markets (tens of millions of units/year in autos alone).

---

So yes, it’s absolutely worth noting Arm’s “optional AI accelerator” comment. It’s practically an invitation for Akida/TENNs to sit alongside Ethos inside Zena CSS.

 

[Gies]

So yes to my humble opinion we’re with Merc for sure

 

[TheDrooben]

Hi Larry,

I’ve found ChatGPT to be extremely useful in the right situations.

But, my advice is - if you don’t understand the subject matter well enough, don’t prompt ChatGPT willy-nilly. Otherwise, you’ll have no way of knowing whether what it’s telling

Hi Larry,

I’ve found ChatGPT to be extremely useful in the right situations.

But, my advice is - if you don’t understand the subject matter well enough, don’t prompt ChatGPT willy-nilly. Otherwise, you’ll have no way of knowing whether what it’s telling you is actually true.

Wow...... while I absolutely respect your posts throughout the years Bravo IMO that is absolute rubbish. Let's face it....if you post something based on ChatGPT or Grok without evidence to back it up then you might as well be a Scientologist or any other "religion"

 

[Bravo]

Wow...... while I absolutely respect your posts throughout the years Bravo IMO that is absolute rubbish. Let's face it....if you post something based on ChatGPT or Grok without evidence to back it up then you might as well be a Scientolotogist or any other "religion"
View attachment 90768

Time will tell...

 

[Fullmoonfever]

View attachment 90772

Thanks @Gies

Tired eyes and skimming...didn't notice that slide.

Well across us then still.

 

[Frangipani]

Can’t be long now , now that Steve Thorne has deleted BrainChip as his employer in his LinkedIn profile’s header and updated the Work Experience section to now read “Nov 2023 - Sep 2025 1 yr 11 mos” relating to his position as BrainChip Vice President of Sales - together a clear indication he is leaving / has left our company…

Steve Thorne - CelLink Corporation | LinkedIn

Senior Sales & Marketing Executive with 30 years' experience in AI and Data Center… · Experience: CelLink Corporation · Location: San Jose · 500+ connections on LinkedIn. View Steve Thorne’s profile on LinkedIn, a professional community of 1 billion members.

www.linkedin.com

View attachment 90623


Well, the signs were there:





And I also took this to be a hint:

Why would our new VP of Business Development James Shields - rather than Steve Thorne himself - have posted on LinkedIn “I’m hiring!”, despite both advertised job positions “reporting directly to the VP of Global Sales”?

LinkedIn Login, Sign in | LinkedIn

Login to LinkedIn to keep in touch with people you know, share ideas, and build your career.

www.linkedin.com

View attachment 90624


View attachment 90625 View attachment 90626


Also, it struck me as very weird that our company has to date not yet officially welcomed the new VP of Business Development, although he himself updated his LinkedIn profile three weeks ago and shortly after shared the news with his network…





Neither has our company commented on the recent hiring of Dan Ganousis as Director of Sales, yet. I discovered his new role by chance when checking out his profile on LinkedIn after I had noticed him liking BrainChip posts…

Now that the long US Labour Day weekend is over, can we finally (!) expect a company update?!

Further to Wednesday’s post above .

Looks as if we will soon see a LinkedIn post by Steve Thorne saying: “I’m happy to share that I am starting a new position as CCO with…”

Yesterday, he edited his job title in his LinkedIn profile’s header from “VP of Sales” to “Chief Commercial Officer”, but so far hasn’t updated the Work Experience section.

Steve Thorne - CelLink Corporation | LinkedIn

Senior Sales & Marketing Executive with 30 years' experience in AI and Data Center… · Experience: CelLink Corporation · Location: San Jose · 500+ connections on LinkedIn. View Steve Thorne’s profile on LinkedIn, a professional community of 1 billion members.

www.linkedin.com

Just a wild guess - could his new employer by any chance be Hailo, the Tel Aviv-headquartered Edge AI company our new VP of Business Development James Shields just left?

The reason for my speculation is this recent exchange of comments with Max Glover, Hailo’s new Chief Revenue Officer:

Hailo is pleased to announce the appointment of Max Glover as the company’s Chief Revenue Officer (CRO). With over two decades of experience leading global sales and go-to-market organizations in… | Hailo | 248 comments

Hailo is pleased to announce the appointment of Max Glover as the company’s Chief Revenue Officer (CRO). With over two decades of experience leading global sales and go-to-market organizations in the semiconductor and technology sectors, Max joins Hailo at a pivotal moment as the company scales...

www.linkedin.com

“Thanks Steven- let’s compare notes soon!”

Hmmm, they obviously know each other, which is not surprising given that they were both concurrently working for Intel for many years, but the above remark by Max Glover struck me as odd.

So I checked out Hailo’s website, and it turns out they currently don’t have a CCO, plus a comparison of what their website looks like as of today with a screenshot taken on 4 August reveals that their Munich-based VP of Sales, Jan-Friso Blacquière, must have meanwhile left the company (although this is not apparent from his LinkedIn profile).

This is what Hailo’s Leadership Team looks like right now:

The Leading AI Chip Company For Edge Devices | Hailo

As a leading AI chip company, our smart edge solution bridges the gap between existing and future AI technologies and the edge computing.

hailo.ai

And this is what it looked like a month ago:

The Leading AI Chip Company For Edge Devices | Hailo

As a leading AI chip company, our smart edge solution bridges the gap between existing and future AI technologies and the edge computing.

web.archive.org

So they are sure to appoint another new VP of Sales or a CCO soon. We’ll see.

 

[Diogenese]

I don't know if Ghost Bat includes Akida, but the latest version has "improvements (which) included developments to the wing, GPS, sensors, mission payloads, integrated combat system and the autonomous systems."



Published today 20250905

https://www.msn.com/en-au/technolog...TS&cvid=cd86baf6faf8493e973e1682992a4b1d&ei=9

Hundreds of millions of dollars are set to be poured into Australia's "world leading" drone the federal government says can match anything China produces.

Defence Industry Minister Pat Conroy visited South Australia's top-secret Woomera range on Friday, declaring he was "absolutely satisfied" with Boeing's MQ-28 Ghost Bat drone.

But he wouldn't confirm when the aircraft can expect to enter service.

"I'm not going to get ahead of government decision making, but I will say that so far ... they've been hitting every milestone, and we see a definite role for this in the future mix," Mr Conroy said.

Following the first phase of testing's completion, the government is poised to spend about $500 million more on the next stage of the Ghost Bat's development.

The government is preparing to ramp up manufacturing capacity and buy more aircraft in addition to the 11 already purchased.

It is the first combat aircraft to be designed and made in Australia in 50 years, with $1 billion previously committed to its development.

Able to travel at just below the speed of sound, the drone is designed to serve as a "loyal wingman" to help protect and support crewed aircraft carry out a range of activities, including combat roles.


https://en.wikipedia.org/wiki/Boeing_MQ-28_Ghost_Bat

The RAAF initially planned to buy three Airpower Teaming System (ATS) systems, as part of the Loyal Wingman Advanced Development Program (LWADP).[20] These three drones were built at an automated production line in Melbourne, Victoria.[12][10] The order was increased to six with an A$115 million contract days after the first flight.[21] As of 9 May 2023, the Australian government confirmed its commitment to funding 10 aircraft for the RAAF,[4][22] not including three prototypes that will not be owned by the government or operated by the RAAF. This takes the government's total investment in the Loyal Wingman program to over A$600 million. The uncrewed platforms are scheduled to enter service with the RAAF in 2024-25.[23]

In February 2024, the Australian Government announced a further A$400 million in funding for three more aircraft, built to the new "Block 2" standard.[24][25] The announced improvements included developments to the wing, GPS, sensors, mission payloads, integrated combat system and the autonomous systems.[7][26]

I'm not sure if the following is relevant, but it's not Akida.

This mob have been working on satellite swarms with AuDoD using Jetson:

AkulaTech

https://www.akulatech.com/nexus

​

THE BRAIN​

‍​

Australia's most advanced AI powered Smart Satellite​

Nexus-01 is an Edge-AI payload designed and developed to demonstrate advanced on-orbit AI capabilities. It processes and analyses hyperspectral data from Level 0 to Level 3, and runs multiple AI models in parallel, to extract real-time, actionable intelligence.
Nexus-01 marks a bold step toward truly autonomous, Software Defined Smart Satellites.

​

AI Supercomputer Onboard​

NVIDIA Jetson TX2i with 256 CUDA cores and 1.3TFLOPS - the brain of the mission.

Satellite swarms have a more predictable path than drone swarms, so Akula may not be working on Ghost Bat.

 

[Diogenese]

He is not talking specifically about brn yet but the picture shows AKIDA 1 and AKIDE 2

That is the first time I've seen them refer to Akida2, and confirms that they have continued to keep Akida in the loop. As part of the early adopters club, they would also be monitoring Akida 3 and GenAI.

 

[TheDrooben]

Time will tell...

Would you rather a ChaptGPT thread or a @Frangipani thread ......I know what I would prefer.......selfless people sharing their own amazing in-depth validated research or a thread based on ChatGPT or other baseless rubbish.....????

 

[Bravo]

Wow...... while I absolutely respect your posts throughout the years Bravo IMO that is absolute rubbish. Let's face it....if you post something based on ChatGPT or Grok without evidence to back it up then you might as well be a Scientolotogist or any other "religion"
View attachment 90768

I don't beleive I was promoting something akin to Scientology or any other organised religion like that.

I was just making a point that I thought I could prompt ChatGPT with a level of confidence, that others might not have, by virtue of the fact that I've been around so long and have done so much due diligence and research on it - perhaps in a much more OCD way than others might.

What happened after we all found out that Arm’s Cortex-M85 and Akida could integrate successfully?

Nothing...

Why is that?...